On the Sunday night of this year's Reality Escape Convention in Laval, Gmail blocked the accounts that were due to send a newsletter to the game's players the next morning. On Monday, from the stage, attendees were told that a company called Impish Posters and Marketing had wanted to send its newsletter and was having trouble with its email server.
Both statements describe the same event. The second one was available because the company had been written, before the convention opened, as the sort of outfit whose email would fail.
Mark Larson, who built the convention's game with David Spira and Theresa Wagner, has published two accounts of it on Room Escape Artist: a walkthrough on 4 October and a piece on what he calls adaptive design on 11 October. I was not there and have not played it. Everything below comes from those two articles, which means it comes from the designer.
A sponsor designed to be late
The game ran in the gaps of a two-day convention, around the talks and the exhibit hall. Its fiction was a sponsor that had failed to get itself to the event. Impish had a booth of strewn cardboard boxes, a half-built website, patchy social media accounts and a flyer that said very little. Read the name quickly and it becomes "imp posters", which is the joke and also the confession.
The puzzles lived on those surfaces. A flyer handed out at the first speaker session sent players to the signs outside the convention rooms. An email to a staff member called Riley drew an automatic reply pointing to the booth, where a keypad and a hidden receipt printer gave each player one of four QR codes and one of four "lucky codes". The next morning's message carried one of four videos. Nobody held enough to finish alone, which was the intention. Larson lists the team's three goals as getting attendees to explore the venue, getting them to work with each other, and spending as little staff effort as possible. More than a hundred people finished.
He also says something that most designers would prefer not to print. A game of this size cannot be playtested in any useful way, and the team could not watch everyone who played it.
The backup was built for someone else
The piece of the design I find most instructive is the one that rescued the Monday.
The newsletter's contents had already been copied onto the Impish website. That was done for a different reason. The team knew that some players would reach that stage after the first day's material had gone out, and wanted them to pass straight through without waiting on an inbox. When the accounts were blocked, the same copy served everybody. The stage announcement, delivered by Lisa Spira, sent players back to the QR codes they already held.
Larson's first rule in the article is to know why each element of the game exists, so that a change made in a hurry repairs something and breaks nothing else. In 1929 G. K. Chesterton put the same thought into a parable about a fence across a road: the reformer who cannot say what the fence is for should not be allowed to pull it down. Chesterton was arguing about institutions. Larson applies it to his own work, a few days before opening, with the venue's signage turning out to be unavailable for Saturday check-in and the opening puzzle depending on that signage. He moved the flyer to Sunday morning and accepted that players would have less time to wander the building, because he could say which of his three goals the flyer served and how much of it he was giving up.
A duplicate made for late arrivals that turns out to cover an outage is the kind of redundancy I keep noticing in puzzles themselves, where a crossing letter confirms an answer it was never placed there to confirm.
What the cover costs
On Saturday I wrote about an Oxford study of how people read a wrong-answer signal. A solver whose answer is refused has to decide whether the fault lies with them or with the checker. Impish adds a third candidate. A player with an empty inbox on Monday morning could blame themselves, could suspect the game had broken, or could conclude that the company was being itself.
Larson's claim is that the third reading held. In his words, "players never lost confidence in the game." I take that as a designer's report from a team that, by his own account, could not watch everyone who played.
The cost is visible in a detail he includes against his own interest. A couple of players had received the newsletter after all, heard from the stage that it had failed, and were confused. For them the cover story contradicted the evidence in their own inbox. Fiction can absorb a fault only for the people who experienced the fault.
There is a quieter cost as well. The first puzzle resolved to the phrase ASK 4 HELP, and convention staff were told to answer any question about it by pointing players toward other attendees. That teaches a useful lesson early: if you are confused, somebody else is holding the missing piece. It is the same lesson a player would need to unlearn in order to recognise a real breakage. In this game a designed gap and an accidental one were made to look alike on purpose, and the players were trained to respond to both in the same way, by finding another person.
That seems to me a fair trade at a convention full of people who build escape rooms for a living. I am less sure how far it travels. A room with one team inside it and a clock running has nobody else to ask. If its story explains every stuck drawer as part of the world, what is left for the players to tell them that the drawer is simply broken?